| 학회 |
한국화학공학회 |
| 학술대회 |
2003년 가을 (10/24 ~ 10/25, 한양대학교) |
| 권호 |
9권 2호, p.1736 |
| 발표분야 |
공정시스템 |
| 제목 |
Batch process monitoring using MPCA based on variable-wise unfolding and time varying score covariance |
| 초록 |
Multiway principal component analysis (MPCA) has been widely used to monitor batch processes. However, it has such shortcomings for on-line batch monitoring that the future behavior of the new batch should be inferred somehow and all batch length should be equalized. A new statistical batch monitoring approach is proposed to overcome the drawbacks of MPCA and obtain better monitoring performance. After eliminating the batch trajectory and scaling the variables at each time, the batch data is rearranged to a form of variable-wise matrix. Then, the covariance matrices of scores at each time during a batch are calculated for the extracted scores to incorporate the covariance variations at each time. These procedures don’t have to anticipate the future values while the dynamic relations are preserved. The proposed method was applied to monitoring of the simulated fed-batch penicillin fermentation. The simulation results clearly show the power and advantages of the proposed method in comparison to MPCA. |
| 저자 |
이종민1, 유창규2, 이인범3
|
| 소속 |
1포항공대 화학공학과, 2Ghent Univ. BIOMATH학과, 3Belgium |
| 키워드 |
Batch monitoring;MPCA;fault detection;fault identification |
| E-Mail |
|
| 원문파일 |
초록 보기 |
